LADI: A Latent Discriminant Model for Analyzing Marketing Research Data

Abstract

A general, flexible LAtent DIscriminant model is described. LADI is a model-based clustering procedure, derived from a specific conceptualization in which the discrimination problem is viewed in a latent mixture context. The basic model yields maximum likelihood (ML) estimates of mixing parameters and structural parameters that define the latent clusters in terms of the responses to a set of descriptor variables. Among other features, the model accommodates descriptor variables having different scale properties, allows for the investigation of group structure, provides a statistical test of the number of latent clusters to retain, and allows for constraints to be imposed on the solution.